We consistently hear stories about workers resisting AI. About employees worried they’ll be replaced. About organizations struggling with adoption because people just don’t want to change.
But here’s what we’re actually seeing: the resistance isn’t coming from where everyone thinks it is.
Zapier recently surveyed 532 U.S. enterprise leaders at companies with 1,000+ employees, and the results tell a different story than the narrative dominating tech headlines. Turns out, 92% of enterprises are treating AI as a priority. More than half of leaders (56%) consider themselves “enthusiastic champions” of AI adoption, with another 28% cautiously optimistic. Only 4% are actively resistant.
So if everyone’s excited about AI, why isn’t it working?
The answer isn’t about fear. It’s about something far more practical: integration.
The Integration Gap Nobody Talks About
Seventy-eight percent of businesses reported difficulty connecting AI software to existing systems as a challenge. Moreover, over half of the respondents (53%) view the process of integrating AI software as moderately to extremely difficult.
Consider the implications here. They’ve got leadership that trusts the power of AI. They’ve got the teams in place that are ready to leverage it. The problem arises when they try to put it in place. They can’t, not because it isn’t effective, but because it won’t work until it can integrate with the dozens of other systems they’re currently using.
The gap between the desire for AI and its implementation in an enterprise setting is enormous. The key challenges aren’t related to faith in AI’s potential. They cite difficulties in adopting AI systems, a lack of skills, costs, and lock-in.
The issue of integration isn’t as attention-grabbing as that of AI stealing jobs. But it’s the true constraint that prevents most companies from fully realizing AI’s potential.
The Vendor Headache
The integration problem gets worse when you look at the vendor landscape. Almost half of enterprises (45%) cite high costs as a barrier to AI adoption. Another 38% lack trust in vendor security. A third (33%) fear vendor lock-in.
This isn’t paranoia. These are legitimate concerns for organizations that have spent years building complex technology stacks. Introducing AI means introducing yet another vendor, another set of security reviews, and another potential single point of failure.
And because these AI tools often can’t connect seamlessly with existing infrastructure, IT teams end up building custom integrations. That takes time, requires specialized skills, and creates technical debt that will have to be maintained for years.
The Skills Gap Is Real
Speaking of skills: 35% of leaders cite AI skill gaps among employees as a barrier to adoption. But this isn’t just about needing data scientists.
The survey reveals a stark departmental divide. IT is over 10 times more likely to lead AI acceleration than sales, marketing, HR, or customer service teams. Yet IT and the C-suite are also the departments most likely to hinder AI adoption due to infrastructure limitations and approval processes.
That’s a challenging dynamic. The people who understand the business problems AI could solve often lack the technical expertise or confidence to implement it. People with technical expertise are already overwhelmed managing existing systems, addressing security concerns, and figuring out how everything connects.
The Competitive Pressure Is Intense
But the clock keeps ticking. Eighty-one percent of firms feel the pressure to accelerate AI adoption to remain competitive, while 41 percent admit that lagging AI adoption has left them behind.
This creates a set of challenges in itself. Companies are in a rush to implement AI without first laying a foundation for its successful integration. They struggle with disparate systems, annoyed employees, and untold ROI.
Companies that reported slowly adopting AI reported the following consequences: losing ground to competitors (41 percent), forgoing opportunities to improve efficiency/productivity (39 percent), and reduced ROI times (37 percent).
What Actually Works
The way forward isn’t unclear. There are key characteristics of organizations that succeed in scaling AI. They prioritize orchestration over adoption. They ensure the AI applications they develop work seamlessly with each other and with other systems in place. They develop AI-driven workflows across the organization rather than AI-centric prototypes.
The story isn’t about people resisting AI. This is the challenge organizations face in dealing with a complex scenario. They must adopt transformative technology and address security, legacy systems, and teams with varying AI knowledge.
Successful companies democratize AI so it can be leveraged by team members, regardless of technical expertise, while still maintaining IT control over compliance and security. This isn’t about bypassing IT. This is just giving people the opportunity to solve more problems using AI, within the right boundaries.
Beyond the Hype
This is a tipping point moment in enterprise AI. The excitement is genuine. The potential is enormous. But it’s time to move beyond the notion that the greatest barrier to adoption is workforce resistance to change.
The challenge, therefore, is that of integration. And this means integrating AI systems with the reality of the enterprise. That involves costs, security, vendor management, and aligning the technical organization with the business organization. That means developing the infrastructure necessary for AI to live up to its potential.
Those organizations that understand this truism and make the necessary investment will realize the value in AI. Others, chasing the next bright AI technology without addressing the challenges of integrating it, will continue to wonder why returns are not as expected.
The AI revolution isn’t stifled by a generation in fear of the future. The AI revolution is hampered by the challenge of integrating this new technology with the other technologies that have been developed. We can solve the problem when this reality is accepted.
About the Survey
The survey was conducted by Centiment for Zapier between September 19 and September 23, 2025. The results are based on 532 completed surveys. Respondents were screened to be U.S. C-Suite Executives, Presidents, Owners, or Partners at companies with 1,000+ employees. Data is unweighted, and the margin of error is approximately +/-4% for the overall sample with a 95% confidence level. Full survey results: https://zapier.com/blog/ai-resistance-survey/
